The Vocabulary Problem Hiding in Your AI Strategy
In this episode of the Pure Report, we are joined by Melody Zacharias, Technical Evangelist Director at Everpure. We start by digging into Melody's background: decades as a DBA going all the way back to COBOL and DB2, early SQL Server deployments, multiple published books, a longtime Microsoft MVP, and an executive MBA she just finished. We discuss what she took away from that MBA program — unlike a lot of people who treat it as three letters on a business card, she found virtually everything immediately transferable to her actual work.
From there, we dive into the core premise: a lot of AI projects and data consolidation initiatives are stalling, and the teams running them often can't articulate why. Her argument is that the failure isn't necessarily technical — it's linguistic. AI is moving so fast that people are not only unfamiliar with the vocabulary, they're actively misusing it, and they don't understand why it matters to them. We walked through four layers of the same stack, each answering a different strategic question, using a back-to-school analogy she built out. Data primacy is the primer — one shared notebook every classroom works from, instead of a different notebook per room. Data lineage is what goes inside that primer. Basically, the family tree of your data, where it came from, where it's going, and how it connects. Next is Ontology, her favorite, and the layer she says people are forgetting right now, the grammar and spelling: the context that tells a system an order number in your CRM and a number in another system are actually the same thing. And the Enterprise Data Cloud is the school building itself, the structure that holds the students, the classrooms, and the books together in one environment.
We also talk through why this is happening now. The app-centric era of standalone CRM, ERP, and SCM systems built silos that aren't going anywhere. The advantage goes to organizations that can extract value from information regardless of where it sits. That requires context, and context is exactly what AI lacks unless you give it to it. Tribal knowledge lives in people's heads; AI has no tribal knowledge, and no human can manually sift millions of data points to find the connections. We connect this back to a previous episode where we covered the stat that 95% of AI pilots fail, and Melody noted that at events like PASS Summit this conversation is only just starting to surface — teams know their projects failed, but not that the root cause was people not understanding each other. Her closing thought sums up the whole episode: words matter. Melody has a companion blog post on this topic, and she's hosting a community session with Argenis Fernandez of Nike demoing how to discover ontology inside your own organization.
To learn more, visit: https://pure.ai
Check out the new Everpure digital customer community to join the conversation with peers and Everpure experts:
https://purecommunity.purestorage.com/
00:00 Coming Up and Intro
01:56 Melody’s Background with Data
07:48 Language Used Around AI
10:18 Defining Data Primacy
12:55 Data Lineage
14:10 Ontology of Data
19:53 Enterprise Data Cloud
